CI Update

[matekelemen]

Trying to resolve CI failure in #368.

upload-artifact v1 through v3 are deprecated, so I updated it to v4.

[matekelemen]

Ok I see this isn't the only problem. I'll have a look a bit later.

[philbucher]

I think its mainly outdated

Python 3.8 is EOL and thus no longer available, same for some compilers

[matekelemen]

I think its mainly outdated

Python 3.8 is EOL and thus no longer available, same for some compilers

Yep, so I think it's up to you what you want to support specifically. Should I update the CI to some versions I find useful, or will you handle the updates yourself?

[philbucher]

I would appreciate if you could do it

[matekelemen]

There might be a memory corruption somewhere. @philbucher have you seen this one before?

EDIT: Nevermind, I moved a line I shouldn't have by mistake.

[matekelemen]

@philbucher

Apart from setting up Rocky Linux this PR is ready for review, take a peek please.

Changelog

CI

• replace Ubuntu20.04 (EOL) to Ubuntu22.04
• replace the specific version of clang (clang-11) on Ubuntu with the default provided by the system's pacman instead (currently clang-14)
• replace the legacy intel compiler (EOL since 2023) with the new one
• replace python38 (EOL) with python312 and python313
• replace Centos 7 with Rocky Linux 8 (WiP)
• bump actions/checkout from v2 to v4
• bump actions/upload-artifact from v2 to v4
• bump actions/setup-python from v1 to v5
• bump jwlawson/actions-setup-cmake from v1.10 to v2
• bump old CMake test job from v2.8.x to v3.15.x

Tests and Examples

• explicit void as argument for functions that don't take any (required by strict compiler settings)
• remove unnecessary template parameters (required by strict compiler)
• explicitly provide argc and argv for every main function. This one's pretty nasty because C doesn't let me omit the argument names, but then complains about unused parameters. So I had to (void)(argc) in every main function ...

[philbucher]

great work, thanks a lot

minor comments

[matekelemen]

@philbucher this PR is ready for review.

Can you please take a look at the MacOS runs? It failed twice on the same parent test case but on different subcases, then passed on the third try. I doubt it had anything to do with this PR.

[philbucher]

awesome, thanks for the work

Regarding Mac, seems that the local-sockets comm fails sometimes. Its been a while, but last time I was working on it, this feature was still a bit under development and perhaps not fully stable.